Digital system testing, troubleshooting and measurement
Scope of the domain
This domain tests how a digital board is proved working or diagnosed when it is not: fault models, test vectors, design for testability, boundary scan, built-in self-test, and the correct use of oscilloscopes, logic analysers and multimeters.
Core rules
- A stuck-at fault fixes a node at 0 or 1. A bridging fault shorts two nodes; an open leaves a node floating or weakly driven.
- Test vectors apply inputs and compare outputs with the expected truth table. Good vectors detect all modelled faults and avoid redundant tests.
- Scan design turns flip-flops into a shift chain so internal states can be loaded and observed. Boundary scan (JTAG) tests interconnects through a TAP controller and boundary-scan cells.
- BIST generates test patterns on-chip, often with an LFSR, and compresses responses into a signature.
- An oscilloscope measures waveform shape and timing; bandwidth should be at least about five times the highest signal frequency. A logic analyser captures many channels and decodes buses.
- Troubleshoot in order: supply voltage, ground, clock, reset, then signal activity. Divide the circuit in half and compare expected with measured levels.
Worked example
A two-input AND gate should drive an LED high when both inputs are high, but the output stays low. A voltmeter shows 5 V on both inputs and 0 V on the output. Disconnect the gate output from the downstream load and measure again. If the output rises to about 5 V, the gate is probably good and the load or a solder bridge is pulling the node low; if it remains 0 V with both inputs high, the gate is faulty or its ground/power connection is bad. This split test distinguishes a bad driver from a bad load without unsoldering the IC.
Traps that cost marks
Using a multimeter to judge a fast clock or a brief glitch; using a scope probe with a long ground lead and misreading ringing; assuming an IC is faulty before checking power, ground, clock and reset; and ignoring loading by the probe itself.